Artificial intelligence-based facial proportion analysis and clothing matching recommendation method

Abstract

The present invention relates to the technical field of artificial intelligence and image processing. Disclosed is an artificial intelligence-based facial proportion analysis and clothing matching recommendation method, comprising the following steps: facial image acquisition and preprocessing, facial feature extraction, facial proportion calculation, and clothing matching recommendation. Compared with the prior art, the present invention has the advantages of high degree of personalization, high accuracy, ease of operation, and strong optimizability.

Description

An AI-based method for facial proportion analysis and clothing matching recommendation Technical Field

[0001] This invention relates to the fields of artificial intelligence and image processing technology, specifically to an artificial intelligence-based method for facial proportion analysis and clothing matching recommendation. Background Technology

[0002] With the booming development of e-commerce, especially the rise of apparel e-commerce, users face an increasing number of choices when purchasing clothing. However, how to quickly and accurately select clothing that matches their own characteristics has become a major challenge for users. Traditional clothing recommendation systems mainly rely on data such as users' purchase history and browsing records for recommendations, but this method often ignores users' personalized characteristics and aesthetic preferences, resulting in low accuracy in recommendations.

[0003] In recent years, the rapid development of artificial intelligence technology has provided new solutions for clothing recommendation systems. In particular, recommendation methods based on facial feature analysis can more accurately grasp users' personalized needs and aesthetic preferences by analyzing their facial features, thereby improving the accuracy and personalization of recommendations.

[0004] Summary of the Invention

[0005] The technical problem this invention aims to solve is to provide a more accurate and personalized clothing recommendation method. Specifically, this invention aims to extract a user's facial features and calculate their proportional relationships through deep analysis and processing of facial images, and then recommend suitable clothing to the user based on these proportional relationships. This method can fully consider the user's facial features and aesthetic preferences, achieving a personalized recommendation effect and improving the user's shopping experience and efficiency.

[0006] To solve the above-mentioned technical problems, the technical solution provided by the present invention is: a method for facial proportion analysis and clothing matching recommendation based on artificial intelligence, comprising the following steps:

[0007] Step 1: Facial Image Acquisition and Preprocessing: Users capture facial images using mobile devices or dedicated cameras and upload them to the system. The system preprocesses the images, including noise reduction, enhancement, and cropping, to ensure that the image quality meets the requirements of subsequent analysis.

[0008] Step 2, Facial Feature Extraction: Deep learning algorithms are used to extract features from the preprocessed facial images. Facial features include eyes, eyebrows, nose, and mouth. For each key feature, the system further extracts its position and size information to form a facial feature dataset.

[0009] Step 3: Facial Proportion Calculation: Based on the extracted facial feature dataset, the system calculates the proportional relationships between key facial features. These proportional relationships reflect the user's facial features and aesthetic preferences, and are an important basis for subsequent clothing matching recommendations.

[0010] Step 4, Clothing Matching and Recommendation: Based on the calculated facial proportions, the system divides users into different facial feature categories, including the most beautiful, sexy, and elegant categories. For each category, the system presets a series of clothing styles, colors, and material attributes that match that category. Through the matching algorithm, the system retrieves clothing from the clothing database that matches the user's facial feature category and sorts and recommends them according to a certain priority.

[0011] Step 5: Display and Optimization of Recommended Results: The system will display the recommended clothing to the user in the form of a list or images, and provide detailed clothing information and purchase links. Users can filter and adjust the recommended results according to their personal preferences and needs, such as modifying clothing style, color and other attributes.

[0012] The advantages of this invention compared to existing technologies are: 1) High degree of personalization: By analyzing the proportional relationships of facial features, it is possible to more accurately grasp the user's personalized needs and aesthetic preferences, thereby achieving personalized recommendation effects. This not only improves the accuracy of recommendations but also meets the user's personalized needs;

[0013] 2) High accuracy: This method utilizes deep learning algorithms to extract features and calculate proportions in facial images, accurately extracting user facial features and calculating their proportional relationships. This ensures the accuracy and reliability of subsequent recommendations;

[0014] 3) Simple operation: Users only need to take and upload a facial image, and the system can automatically complete steps such as facial feature extraction, proportion calculation, and clothing recommendation, without complicated operations. This lowers the user threshold and improves the ease of use of the system;

[0015] 4) High Optimization Potential: The system can continuously optimize its recommendation algorithm based on user feedback and behavioral data, improving the intelligence and personalization of recommendations. This ensures the system can adapt to the needs and changes of different users, maintaining long-term competitiveness.

[0016] 5) Broad application prospects: This method is not only applicable to the apparel e-commerce field, but can also be extended to other fields that require personalized recommendations, such as cosmetics and jewelry, which provides new ideas and methods for the development of related fields.

[0017] Preferably, in step three, the ratio of the distance between the eyes to the distance from the eyebrows to the tip of the nose is (y1 / x1), and the ratio of the width of the lips to the distance from the tip of the nose to the tip of the chin is (y2 / x2).

[0018] Preferably, the system continuously optimizes the recommendation algorithm based on user feedback and behavioral data to improve the accuracy and personalization of recommendations.

[0019] Preferably, the method is based on a dedicated personalized clothing recommendation system, which includes a facial feature extraction module, a facial proportion calculation module, a clothing recommendation module, and a user feedback module. The facial feature extraction module is used to extract key facial features of the user, the facial proportion calculation module is used to calculate the facial feature proportion and classify the user's facial features according to the proportion, the clothing recommendation module is used to recommend suitable clothing according to the category to which the user's facial features belong, and the user feedback module is used to receive user feedback on the recommended clothing and optimize the recommendation algorithm.

[0020] Preferably, the facial feature extraction module extracts key facial features from the facial image uploaded by the user using image processing technology;

[0021] Preferably, the facial proportion calculation module divides the user's facial features into multiple categories based on the ratio of (y1 / x1) and (y2 / x2), including the most beautiful, sexy, elegant, mysterious and friendly categories.

[0022] Preferably, the clothing recommendation module retrieves and recommends clothing from the clothing database that matches the user's facial feature category.

[0023] Preferably, the system also includes the ability to integrate with existing facial recognition technology to achieve automatic identification and recommendation. Attached Figure Description

[0024] Figure 1 is the face diagram of this invention patent.

[0025] Figure 2 is the second face diagram of this invention patent.

[0026] Figure 3 is a schematic diagram of the golden ratio of this invention patent. Detailed Implementation

[0027] Example 1

[0028] In specific implementation, this invention includes the following detailed steps:

[0029] I. Facial Image Acquisition and Preprocessing

[0030] Users capture images of their faces using mobile devices or dedicated cameras and upload them to the system. The system automatically preprocesses the images, including noise reduction, image enhancement (such as contrast boosting and sharpening), and cropping, ensuring that the image contains only the user's face and removing background noise.

[0031] II. Facial Feature Extraction

[0032] The system utilizes deep learning algorithms, such as convolutional neural networks (CNNs), to extract features from preprocessed facial images. The extracted facial features include information such as the shape, size, and location of key areas like the eyes, eyebrows, nose, and mouth.

[0033] III. Facial Proportion Calculation

[0034] Based on the extracted facial feature information, the system calculates the proportional relationships between key facial features, such as the ratio of the distance between the eyes to the distance from the eyebrows to the tip of the nose (y1 / x1), and the ratio of the width of the lips to the distance from the tip of the nose to the tip of the chin (y2 / x2). These proportional relationships will serve as an important basis for subsequent clothing matching recommendations.

[0035] IV. Clothing Matching Recommendations

[0036] Based on the calculated facial proportions, the system categorizes users into different facial feature categories, such as most beautiful, sexy, and elegant. For each category, the system pre-sets a series of clothing styles, colors, and material attributes that match that category. The system uses a matching algorithm to retrieve clothing from the clothing database that matches the user's facial feature category, and then sorts and recommends clothing according to a certain priority (such as popularity, rating, etc.).

[0037] V. Display and Optimization of Recommendation Results

[0038] The system displays recommended clothing items to users in the form of a list or images, providing detailed clothing information and purchase links. Users can filter and adjust the recommendations according to their personal preferences and needs, such as modifying clothing styles, colors, and other attributes. Based on user feedback and behavioral data, such as clicks, purchases, and ratings, the system continuously optimizes the recommendation algorithm to improve the accuracy and personalization of the recommendations.

[0039] Example 2

[0040] Based on Example 1, the specific development process of this solution includes:

[0041] 1) Hardware Environment: The system requires high-performance servers and storage devices to process large amounts of user data and run deep learning algorithms. Simultaneously, the user end needs mobile devices or dedicated cameras that support photo taking and uploading functions.

[0042] 2) Software Development: Develop backend server programs using programming languages ​​such as Python. These programs are responsible for receiving user-uploaded images, performing image processing and facial feature extraction, calculating facial proportions, and providing clothing matching recommendations. The frontend uses HTML, CSS, JavaScript, and other technologies to develop the user interface, which displays recommendation results and provides user interaction functions.

[0043] 3) Database Design: Establish a user information database and a clothing database. The user information database stores users' facial features, aesthetic preferences, and other data; the clothing database stores detailed information on the styles, colors, materials, and corresponding image resources of various garments.

[0044] 4) Algorithm Implementation: Facial feature extraction and proportion calculation algorithms are implemented using deep learning frameworks (such as TensorFlow, PyTorch, etc.). Appropriate matching algorithms (such as cosine similarity algorithm, K-nearest neighbor algorithm, etc.) are used for clothing matching and recommendation. Simultaneously, combining user feedback and behavioral data, machine learning algorithms (such as gradient descent algorithm, logistic regression algorithm, etc.) are used to continuously optimize the recommendation algorithm.

[0045] 5) Testing and Optimization: Thorough testing will be conducted during development to ensure system stability and accuracy. Simultaneously, the system will be continuously optimized and improved based on user feedback and usage, enhancing user experience and satisfaction.

[0046] Example 3

[0047] Based on Embodiment 2, the system functional features accompanying this solution include:

[0048] I. Personalized Recommendation Function

[0049] The core function of this system is to provide personalized clothing recommendations based on the user's facial proportions and aesthetic preferences. Through deep learning algorithms, the system precisely analyzes the user's facial images, extracting key facial features and calculating their proportions. Based on this facial proportion information, the system can categorize users into different facial feature classes and pre-set matching clothing styles, colors, and material attributes for each class. This personalized recommendation method ensures that users receive clothing recommendations that match their facial features and aesthetic preferences, thereby improving shopping satisfaction.

[0050] II. Intelligent Matching Algorithm

[0051] The system employs advanced intelligent matching algorithms, matching users based on their facial proportions and preset clothing attributes. These algorithms consider not only clothing style, color, and material attributes, but also user behavior data such as purchase history and browsing records, as well as user feedback on the recommendations, continuously optimizing the results. Through these intelligent matching algorithms, the system can provide users with more accurate clothing recommendations tailored to their needs, improving shopping efficiency and experience.

[0052] III. Real-time updates and adjustments

[0053] The system can update information in its clothing database in real time, including newly added garments, out-of-season items, and inventory changes. Simultaneously, the system can adjust its recommendation algorithm and results in real time based on user feedback and behavioral data. This real-time updating and adjustment capability allows the system to maintain close contact with the market and users, ensuring the timeliness and accuracy of recommendations.

[0054] IV. User Interaction and Feedback

[0055] The system offers rich user interaction features, allowing users to filter, adjust, and provide feedback on recommendations. Users can modify attributes such as clothing style and color according to their preferences and needs to obtain recommendations that better suit their requirements. Users can also rate and comment on the recommendations, providing valuable feedback to the system. This feedback will be used by the system to optimize the recommendation algorithm and improve recommendation quality.

[0056] V. Multi-platform support

[0057] This system supports multiple platforms, including web and mobile applications. Users can use the system on any platform according to their preferences and habits. This multi-platform support makes the system more flexible and convenient, meeting the needs of different users.

[0058] VI. Privacy Protection and Security

[0059] The system strictly protects user privacy and information security. During facial image processing and facial feature extraction, the system de-identifies user facial images to ensure user privacy is not compromised. Simultaneously, the system employs advanced encryption technology and security measures to guarantee the security and integrity of user information. Users can confidently enjoy personalized clothing recommendation services when using this system.

[0060] In designing this scheme, we are based on the concept of double cross face classification method. This concept was proposed by its creator, Chen Yuren. It is a facial classification concept based on his experience of interviewing more than 10,000 people. The face is divided into two parts, as shown in Figure 1.

[0061] The study also found that this concept can be quantified:

[0062] 1) The distance between the outer edges of each eye is x1, and the distance between the eyebrow and the tip of the nose is y1;

[0063] 2) If the width of the lips is quantized as x2, the distance from the tip of the nose to the tip of the chin is quantized as y2;

[0064] Therefore: y1 / x1 ≒ 1.618

[0065] y² / x²≒1.618 can be defined as the most beautiful facial features.

[0066] As shown in Figure 3, the golden ratio diagram, strictly speaking, is 1:168, which is the so-called golden ratio. This law describes a kind of beauty that often appears in nature. One explanation seems to be that it is "a proportion in the DNA of nature".

[0067] Based on the aforementioned existing technology, the applicant has established the following standards for this solution:

[0068] 1) A sexy ratio (y² / x² > 1.618) adds sex appeal to beauty;

[0069] 2)chic y1 / x1<1.618 katsu y2 / x2<1.168 elegant, graceful;

[0070] 3) fantasy y1 / x>1.618かつy2 / x2<1.168 mysterious charm;

[0071] 4) casual y1 / x<1.618 katsu y2 / x2>1.168 familiar, neighborly relations;

[0072] 5)kawaii y1 / x>1.618 cute, childlike.

[0073] These five categories can result in six different facial classifications. A key feature of this classification method is that all classifications are expressed in proportion, regardless of whether the person is Western or Asian, has large or small eyes, or a high or low nose bridge. Once each classification is determined, an artificial intelligence or program is developed to automatically recommend the most suitable clothing and even list brand names. Facial classification uses artificial intelligence.

[0074] Figure 2 is a schematic diagram of Microsoft Face Cognitive Services. ChatGPT now also supports face recognition application programming interfaces (APIs), and other solutions are also available. Generally, API keys and parameters are obtained through programming code (such as Python). In a practical example, uploading a face photo activates the program. The program then classifies the face using a double-cross face classification method, recommends clothing suitable for the face pattern, and sometimes even displays images or recommends clothing brands.

[0075] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A method for facial proportion analysis and clothing matching recommendation based on artificial intelligence, characterized in that... Includes the following steps: Step 1: Facial Image Acquisition and Preprocessing: Users capture facial images using mobile devices or dedicated cameras and upload them to the system. The system preprocesses the images, including noise reduction, enhancement, and cropping, to ensure that the image quality meets the requirements of subsequent analysis. Step 2, Facial Feature Extraction: Deep learning algorithms are used to extract features from the preprocessed facial images. Facial features include eyes, eyebrows, nose, and mouth. For each key feature, the system further extracts its position and size information to form a facial feature dataset. Step 3: Facial Proportion Calculation: Based on the extracted facial feature dataset, the system calculates the proportional relationships between key facial features. These proportional relationships reflect the user's facial features and aesthetic preferences, and are an important basis for subsequent clothing matching recommendations. Step 4, Clothing Matching and Recommendation: Based on the calculated facial proportions, the system divides users into different facial feature categories, including the most beautiful, sexy, and elegant categories. For each category, the system presets a series of clothing styles, colors, and material attributes that match that category. Through the matching algorithm, the system retrieves clothing from the clothing database that matches the user's facial feature category and sorts and recommends them according to a certain priority. Step 5: Display and Optimization of Recommended Results: The system will display the recommended clothing to the user in the form of a list or images, and provide detailed clothing information and purchase links. Users can filter and adjust the recommended results according to their personal preferences and needs, such as modifying clothing style, color and other attributes.

2. The method for facial proportion analysis and clothing matching recommendation based on artificial intelligence according to claim 1, characterized in that: In step three, the ratio of the distance between the eyes to the distance from the eyebrows to the tip of the nose is (y1 / x1), and the ratio of the width of the lips to the distance from the tip of the nose to the tip of the chin is (y2 / x2).

3. The method for facial proportion analysis and clothing matching recommendation based on artificial intelligence according to claim 1, characterized in that: The system continuously optimizes its recommendation algorithm based on user feedback and behavioral data, thereby improving the accuracy and personalization of recommendations.

4. The method for facial proportion analysis and clothing matching recommendation based on artificial intelligence according to claim 1, characterized in that: The method is based on a dedicated personalized clothing recommendation system, which includes a facial feature extraction module, a facial proportion calculation module, a clothing recommendation module, and a user feedback module. The facial feature extraction module is used to extract key facial features of the user, the facial proportion calculation module is used to calculate the proportion of facial features and classify the user's facial features according to the proportion, the clothing recommendation module is used to recommend suitable clothing according to the category to which the user's facial features belong, and the user feedback module is used to receive user feedback on the recommended clothing and optimize the recommendation algorithm.

5. The method for facial proportion analysis and clothing matching recommendation based on artificial intelligence according to claim 4, characterized in that: The facial feature extraction module extracts key facial features from user-uploaded facial images using image processing technology.

6. The method for facial proportion analysis and clothing matching recommendation based on artificial intelligence according to claim 4, characterized in that: The facial proportion calculation module categorizes user facial features into multiple categories based on the ratio of (y1 / x1) and (y2 / x2), including the most beautiful, sexy, elegant, mysterious, and friendly categories.

7. The method for facial proportion analysis and clothing matching recommendation based on artificial intelligence according to claim 4, characterized in that: The clothing recommendation module retrieves and recommends clothing from the clothing database that matches the user's facial feature category.

8. The method for facial proportion analysis and clothing matching recommendation based on artificial intelligence according to claim 4, characterized in that: The system also includes the ability to integrate with existing facial recognition technology to enable automatic identification and recommendations.

Images